Find-Your-Level Placement Quiz Skill: How It Works in the AI Engineering from Scratch Curriculum

The find-your-level placement quiz skill is an interactive 10-question assessment that evaluates five core knowledge areas to determine the optimal starting phase within the 20-phase AI Engineering from Scratch curriculum.

The find-your-level placement quiz skill is a declarative Claude "code skill" included in the rohitg00/ai-engineering-from-scratch repository. Designed to personalize the learning path across 260 lessons, this skill orchestrates a structured dialogue to benchmark a learner's existing expertise in mathematics, classical machine learning, and deep learning.

How the Assessment Flow Works

The skill follows a rigid procedural sequence defined in .claude/skills/find-your-level/SKILL.md. When invoked, it executes a five-round evaluation that progressively scores the learner without revealing correct answers until the final stage.

Quiz Structure and Format

The assessment presents 5 distinct knowledge areas, with 2 multiple-choice questions per area, totaling 10 questions. The areas covered include Math & Statistics, Classical ML, and Deep Learning, among others. Each round focuses on a single area, allowing the skill to calculate partial scores incrementally.

The skill uses the AskUserQuestion tool to present each question and capture user input. After every pair of questions, it reports the area score (e.g., "Math & Statistics: 2/2") before proceeding to the next knowledge domain.

Scoring and Assessment Logic

Scoring is binary: each correct answer awards 1 point, while incorrect or blank responses receive 0 points. This produces an aggregate score ranging from 0 to 10. The skill maintains running totals internally but only provides granular feedback per area to prevent gaming the assessment.

Unlike adaptive quizzes, this skill uses a fixed question set with deterministic correct answers encoded directly in the skill definition file.

Phase Mapping and Recommendations

Upon completion, the total score maps to a specific curriculum phase (1 through 20) using a linear interpolation logic. According to the source specification, the mapping follows this pattern:

• 0–2 points: Phase 1 (foundational)
• 3–4 points: Phase 2
• ...
• 9–10 points: Phase 20 (advanced)

The skill returns a personalized learning path that includes the suggested starting phase and estimated study hours required to reach subsequent milestones.

Invoking the Skill in Claude and Compatible Agents

You can trigger the assessment in any agent that has the curriculum skills installed, including Claude, Cursor, Codex, OpenClaw, and Hermes.

Command Syntax

Use the slash command:

/find-your-level

Alternatively, the skill responds to natural language trigger phrases such as "where should I start", "find my level", or "what do I know" when detected by the agent's intent classification system.

Sample Interaction

Assistant: Hello! I'll run a quick placement quiz to find the best place for you to start.

Assistant: **Round 1 – Math & Statistics**

Assistant: Q1. You have two vectors, a = [1, 2, 3] and b = [4, 5, 6]. What is their dot product?
User: B
Assistant: Correct! (32)

Assistant: Q2. A fair coin is flipped 3 times. What is the probability of getting exactly 2 heads?
User: B
Assistant: Correct! (3/8)

Assistant: Math & Statistics: 2/2
...

The skill aggregates these round-by-round scores before generating the final curriculum placement.

Technical Implementation and Source Files

The entire skill logic resides in a markdown-based skill definition file rather than imperative code. This declarative approach allows the Claude runtime to interpret the quiz flow, correct answers, and interaction rules.

Key File Locations

• .claude/skills/find-your-level/SKILL.md: Contains the complete skill definition, including question text, answer keys, scoring rules, and procedural instructions (e.g., "use AskUserQuestion to ask each question", "score after each round", "explain answers only at the end").

• site/lesson.html (line ~3338): Provides UI hints directing learners to run /find-your-level in supported agents.

• CHANGELOG.md: Documents the version history and addition of the placement-quiz skill to the curriculum ecosystem.

Internal Scoring Algorithm

While the skill definition handles the dialogue state, the underlying scoring logic follows this pseudo-code pattern as implemented in the AI Engineering from Scratch runtime:

def map_score_to_phase(score):
    # 0-2 → Phase 1, 3-4 → Phase 2, … 9-10 → Phase 20

    phase = min(20, max(1, (score // 1) + 1))
    return phase

This ensures every learner receives a deterministic placement based strictly on their 0–10 performance metric.

Summary

• The find-your-level placement quiz skill is a 10-question diagnostic tool defined in .claude/skills/find-your-level/SKILL.md that assesses proficiency across five AI engineering domains.
• It uses binary scoring (0 or 1 per question) to generate a 0–10 aggregate score that maps directly to one of 20 curriculum phases.
• The skill is invoked via /find-your-level or natural language triggers and uses the AskUserQuestion tool to manage the interactive dialogue.
• Feedback is provided incrementally by knowledge area, but detailed answer explanations are deferred until the quiz concludes to maintain assessment integrity.

Frequently Asked Questions

How many questions are in the find-your-level placement quiz?

The quiz contains 10 multiple-choice questions total—2 questions for each of the 5 knowledge areas (Math & Statistics, Classical ML, Deep Learning, etc.). This structure allows the skill to assess breadth across fundamental domains while keeping the total assessment time concise.

Where is the find-your-level skill defined in the repository?

The skill definition lives at .claude/skills/find-your-level/SKILL.md in the rohitg00/ai-engineering-from-scratch repository. This markdown file contains the declarative instructions that tell the Claude runtime how to present questions, validate answers, calculate scores, and map results to curriculum phases.

Can I use the find-your-level skill outside of Claude?

Yes. The skill works in any agent environment that has installed the AI Engineering from Scratch curriculum skills, including Cursor, Codex, OpenClaw, and Hermes. You trigger it using the /find-your-level slash command or recognized natural language phrases like "where should I start".

How does the skill determine which curriculum phase to recommend?

The skill converts your raw score (0–10) into a phase (1–20) using a linear mapping function where higher scores correspond to later phases. For example, a score of 0–2 points places you in Phase 1, while 9–10 points places you in Phase 20. The skill then returns a personalized learning path with estimated hours for your recommended starting point.

{
  "mcpServers": {
    "instagit": {
      "command": "npx",
      "args": ["-y", "instagit@latest"]
    }
  }
}